Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <1587361243-83431-1-git-send-email-xiyuyang19@fudan.edu.cn>
In-Reply-To: <1587361243-83431-1-git-send-email-xiyuyang19@fudan.edu.cn>
From:   Ilya Dryomov <idryomov@gmail.com>
Date:   Mon, 20 Apr 2020 10:44:37 +0200
Message-ID: <CAOi1vP_-82vLSG6UeeLToRGjNepE--fnj8tsb4y8e7eAoH9tCA@mail.gmail.com>
Subject: Re: [PATCH] ceph: Fix potential ceph_osd_request refcnt leak
To:     Xiyu Yang <xiyuyang19@fudan.edu.cn>
Cc:     Jeff Layton <jlayton@kernel.org>, Sage Weil <sage@redhat.com>,
        Ceph Development <ceph-devel@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>, yuanxzhang@fudan.edu.cn,
        kjlu@umn.edu, Xin Tan <tanxin.ctf@gmail.com>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, Apr 20, 2020 at 7:41 AM Xiyu Yang <xiyuyang19@fudan.edu.cn> wrote:
>
> ceph_writepages_start() invokes ceph_osdc_start_request(), which
> increases the refcount of the ceph_osd_request object "req" and then
> submit the request.
>
> When ceph_writepages_start() returns or a new object is assigned to
> "req", the original local reference of "req" becomes invalid, so the
> refcount should be decreased to keep refcount balanced.
>
> The reference counting issue happens in a normal path of
> ceph_writepages_start(). Before NULL assigned to "req", the function
> forgets to decrease its refcnt increased by ceph_osdc_start_request()
> and will cause a refcnt leak.
>
> Fix this issue by calling ceph_osdc_put_request() before the original
> object pointed by "req" becomes invalid.
>
> Fixes: 1d3576fd10f0 ("ceph: address space operations")
> Signed-off-by: Xiyu Yang <xiyuyang19@fudan.edu.cn>
> Signed-off-by: Xin Tan <tanxin.ctf@gmail.com>
> ---
>  fs/ceph/addr.c | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
> index 7ab616601141..b02c050a3418 100644
> --- a/fs/ceph/addr.c
> +++ b/fs/ceph/addr.c
> @@ -1126,6 +1126,7 @@ static int ceph_writepages_start(struct address_space *mapping,
>                 req->r_mtime = inode->i_mtime;
>                 rc = ceph_osdc_start_request(&fsc->client->osdc, req, true);
>                 BUG_ON(rc);
> +               ceph_osdc_put_request(req);
>                 req = NULL;
>
>                 wbc->nr_to_write -= i;

Hi Xiyu,

I think this reference is put in writepages_finish().  Putting it in
ceph_writepages_start() will result in freed pointer dereference in
libceph (in __complete_request() or elsewhere).

Thanks,

                Ilya